OSPF Default Routes E1 and E2

[ADNET_NETSUPPORT]

Hi Folks,

Appreciate if you could answer my query regarding the OSPF based default route.

WAN of Router A is connecting to primary DC site and WAN of Router B with DR.
We are also having one remote site with two routers C & D.
Router C is connected to primary DC router A (LAN side) and router D is connected to secondary DR router B (LAN side).

There is OSPF process 99 running in DC and DR and WAN of both routers A and B are participating in the same routing domain.
Same time LAN of routers A and B are configured with another OSPF process 100 (interfaces connecting with C & D) and there is routes redistribution between 99 and 100 and vice versa.
Routers C and D also connected back to back and running OSPF process (100).

In normal situation, DC and DR sites always have the default route with metric-type E1 and same default route is propagated to routers A and B.
Router A has the command ( default-information originate metric-type 1) and router B (default-information originate always)

The preferred path is always (A <--------> C) with default route metric type E1 and it works fine.
But in case of any issues in DC site, DR site default route with metric E2 is preferred and both A and B would have the default route with E2 metric.

Here is the problem:
As route A is configured to send the default route with E1 and router B with default (in this case it would have E2 route) but it doesn't seem to be working
The traffic from Router C is going to router D and causing Asymmetric flow, which we don't want.

Ideally the flow from A to C should always be preferred but it doesn't seem to be, i have done all my checks but can not find out the reason.
appreciate your help for the same.

Regards

[Rolf Fischer]

Hi,

The traffic from Router C is going to router D and causing Asymmetric flow, which we don't want.

Does that mean that in normal operation Router C's default-route is the E2-route originated by Router B?

If so, could you share the output of

• show ip ospf 100 database external 0.0.0.0
• show ip ospf 100 border-routers

from Router C?

Thanks.

[ADNET_NETSUPPORT]

Hi Rolf,

In normal operation, Router C has the default route as E1, as router A would always be injecting default route with metric type E1 to router C, and Router B with default metric to router D as it is received from DC (E1) or DR (E2).

Router A:

router ospf 99
 router-id 10.254.1.157
 redistribute ospf 100 subnets route-map SAMHA_PRIMARY-TO-DC
 network 10.254.1.160 0.0.0.7 area 5
router ospf 100
 ispf
 log-adjacency-changes detail
 area 5 authentication message-digest
 redistribute connected metric-type 1
 redistribute ospf 99 subnets route-map DC-TO-SAMHA-PRIMARY
 network 10.254.1.156 0.0.0.3 area 5
 default-information originate metric-type 1

Router B:

router ospf 99
 router-id 10.252.1.157
 ispf
 log-adjacency-changes detail
 area 5 authentication message-digest
 redistribute ospf 100 subnets route-map SAMHA-SECONDARY-TO-DR
 network 10.252.1.160 0.0.0.7 area 5
 default-information originate
router ospf 100
 ispf
 log-adjacency-changes detail
 area 5 authentication message-digest
 summary-address 10.252.0.0 255.255.0.0
 redistribute connected subnets
 redistribute ospf 99 subnets route-map DR-TO-SAMHA-SECONDARY
 network 10.252.1.156 0.0.0.3 area 5
 default-information originate always
 distance 120

Shortly I would share the output of the commands requested by you.

Regards,

Qamar Zia

[ADNET_NETSUPPORT]

Find the below output, also note that actual process ID is 120, instead of 100:

mss-adnet-samhambzxconnect-ra#show ip ospf 120 database external 0.0.0.0

            OSPF Router with ID (10.254.22.130) (Process ID 120)

                Type-5 AS External Link States

  Routing Bit Set on this LSA in topology Base with MTID 0
  LS age: 482
  Options: (No TOS-capability, DC)
  LS Type: AS External Link
  Link State ID: 0.0.0.0 (External Network Number )
  Advertising Router: 10.254.1.163
  LS Seq Number: 80000089
  Checksum: 0xDDA7
  Length: 36
  Network Mask: /0
        Metric Type: 1 (Comparable directly to link state metric)
        MTID: 0
        Metric: 1
        Forward Address: 0.0.0.0
        External Route Tag: 120

mss-adnet-samhambzxconnect-ra#show ip ospf 120 border-routers          

            OSPF Router with ID (10.254.22.130) (Process ID 120)

                Base Topology (MTID 0)

Internal Router Routing Table
Codes: i - Intra-area route, I - Inter-area route

i 10.254.1.163 [3] via 10.254.1.157, GigabitEthernet0/0.3017, ASBR, Area 5, SPF 43

[Rolf Fischer]

Thanks for the additional information.

Unfortunately I still don't understand some of the details.

Router A, B, C and D share a common OSPF domain, the PID is 100 on routers A and B and 120 on router C, correct?

Router A advertises a conditional default-route (no always-keyword) as Type-1 and router B advertises a default-route unconditionally (always-keyword) as Type-2. So as long as Router A's routing-table has a default-route, the Type-1 default-route should have preference over the Type-2 in this OSPF domain.

Now I expected to see both external LSAs in router C's LSDB but we see only Router A's, so I obviously made some incorrect assumptions.

Could you please elaborate a bit on the OSPF topology (perhaps with a simple diagram?)

Thanks,

Rolf

[ADNET_NETSUPPORT]

The diagram is attached, routers A, B, C, D share common OSPF domain of 120, additionally routers A and B are running OSPF 99 also with DC and DR sites.

 A to C should always be preferred path and default route from A to C is always sent as E1, B to D is secondary path and should be active in case of failure in primary, router B is sending default route to D with default metric received either from DC or DR.

The design works fine till the time DC E1 route is preferred and propagated to routers A & B, as soon as E2 route from DR site takes preference there is a problem, the default route coming from B to D is preferred and secondary path becomes active, ideally A to C path should always be active either E1 or E2 route is received from DC/DR sites.

hope this helps.

Regards,

Qamar Zia

[Rolf Fischer]

Thanks again for the additional information, I believe I understand now.

as soon as E2 route from DR site takes preference there is a problem

When this happens, does Router A have a default route in its routing table? I'm asking because the origination of a default route in PID 100 on router A is conditional.

Regards,

Rolf

[ADNET_NETSUPPORT]

Yes, routers A and B would always have the default routes either E1 or E2, the problem comes when DR E2 is preferred and sent to routers A and B.

Routers A and B advertise it to C and D respectively, ideally in this situation the outbound traffic flow should be same like before C -----> A and than towards DC but it goes like C --->D -----> B which makes asymmetric routing for us as the return traffic comes from A-----> C (also the traffic starts flowing through the secondary link).

Regards,

Qamar Zia

[Rolf Fischer]

Sorry for the late response, I'm on a training this week.

I think in order to make some progress here we really need to take a closer look at the default-route on Router A. Could you please share the output of 'show ip route 0.0.0.0' from Router A in normal operation?

It would also be very helpful to see the output of this command while the problem occurres. I don't know if it's possible to reproduce the problem, but perhaps you can tell us where Router A's defaut-route comes from in this case on the basis of your troubleshooting results? If it's possible to reproduce the problem, I'd also like to see Router A's output of 'show ip ospf database external 0.0.0.0 self-originate'.